University Cobreflex


This University project is due to a fella named Jeff, of a previous Wharfedale W70 project which has three pages in the Wharfedale Index  After modifying his W70D systems, he asked about the University Classic, a horn loaded system useing a C15W 15" woofer, the Cobreflex horn with a T30 driver and the HF206 horn tweeter.

Being familiar with University systems for decades, when asked for advice on those two Classics, I couldn't resist.  Since then, a pair of Cobreflex horns with T30 drivers was purchased along with one HF206 tweeter.  The search for another HF206 is underway.

The ABS Cobreflex is still available and that aroused my curiosity as it's purported to unload at about 250hz.  The 500hz horns I purchased and loaded with a Faital 3FE22 driver performed well but I've always had a desire to have the midrange cover the band from at least 300hz to about 4500hz.  The closest was the Klipschorn, 400hz to 5khz.  A horn going down to 250hz would be within an inch of 24" in length.  Another handicap is finding a driver that could go that low; they are few and far between unless one has several hundred dollars to invest in a pro driver that can handle a few hundred watts.  For home use, a driver capable of handling 20 watts would be more than sufficient.  Given a sensitivity of 106dB, 1w1m, 20 watts would produce an ear shattering level of 119dB

Drivers such as the those made by University will go down to 350hz as will the T25 and the 1824M drivers made by ElectroVoice if fitted to a horn with a flare rate of 250hz  The horns to which I refer are straight axis exponential horns.  The tractrix will be a little shorter for the same cutoff but they are not easy to build unless one has good software as the calculations can be extremely difficult to handle, much more so than those of the exponential horn.  The tractrix equation is unsolvable by conventional algebraic means. If using a cone speaker, not all such speakers are suitable for horn loading.  Also, a phase plug would be necessary if the horn is to operate beyond about 3khz.

Another drawback with horn loaded mids and tweets is the bass section.  Bass horns get vary large, very fast.  The path length of the Klipschorn is about 40 inches and unloads at 47hz, below which it operated as a sealed box.  The Klipsch type corner horns I designed and built in the late seventies unload at 34hz and the path length is 84 inches.  Keep in mind, as big as many consider the Klipschorn, the mouth is only 1/8th the required area since it works in a corner.  A straight axis exponential horn unloading at 34hz would be 13 feet in length, not including the back chamber.  

Anyway, without going into a lengthy diatribe on horns, the Cobreflex is a very practical solution.  Many have shunned it due to its prolific use in public address systems.

That being said, this page is dedicated to Jeff.




The horns just after unpacking

For the curious, the two speakers in the background are Jensen F12N  There are 4 pages on these in the INDEX



One of the units under test




A closeup of the 25v-70V transformer.  The two dangling wires are those of the driver, the transformer being bypassed





The HF-206



One view of the pair with the tweeter time aligned to the mid-range

This made no difference compared to the mid and tweet mouth aligned as in the next photo

The crossover is wired mostly with alligator clips, or are they crocodile clips?



The other side of the assembly

Here can be seen the 50 ohms potentiometers.

the assembly not only looks mechanically unstable, it is.

When moving around the room taking piccies, great care had to be taken so as not to bump it in the slightest.






Frequency responses, 1w1m gated of the Cobreflex horn with each T30 driver, RED and PINK.  The BLACK and GREY are of the ElectroVoice 1824M drivers.   Both the T30 and 1824M drivers are bandwidth rated from 500hz to 5khz.  The 1824M units will be less sharp between 1.2khz and 3khz.




The T30 impedance curves on the Cobreflex horn.  This was not expected so the traces were run on the CLIO, see fig 3





Z BLK=T30a; GREY=T30b

Close inspection of the curves will show the close similarity.  The variations are very likely due to the two 360 degree folds (U-turns) in the Cobreflex.  See figs 4 and 5





Z 1824M-1 GREY=no horn RED=on 500hz horn

The1844M without a horn, BLACK and on the 500hz straight axis horn. RED.  It becomes evident how the horn affects the impedance.

Now look at fig 5 below.





Z GREY=T30; BLK=1824M

The GREY is the impedance of the T30 driver on the Cobreflex. Compare the similarity of this with that shown in fig 2

The BLACK curve is that of the 1824M on the Cobreflex.  Note the dissimilarity between this curve and that of the same driver on the straight axis horn of fig 4 above, the RED curve.

The two bends in the Cobreflex seem to be part of the reason for this; another may be the longer length of the Cobreflex. 

The sharp rise of the 1824M's curve may be due to higher driver impedance without the horn.  see fig 6, RED curve.  A stronger magnetic field in the gap is another factor.





Z RED=1824M2 driver; BLACK=T30A driver

Here's the impedance curves of the drivers only, no horn attached.  The higher resonant peaks of the 1824M may be due to stronger field strengths in the gap.  The lower frequency points of the 1824M may be due to higher diaphragm mass









The following graphs are frequency responses.  All are measured on the Cobreflex horn at 1w1m, on axis and off axis as individually described





1w1m Cobreflex GRN=T30A BLUE=THD; RED=T30B ORN=THD

Comparison of both T30 drivers on axis




1w1m Cobreflex on axis RED=1824M1 GREY=THD; GRN=1824M2 BLUE=THD

Comparison of the T30 and the 1824M on axis




1w1m T30B Cobreflex BLK=on axis; ORN=45 off axix; GREY=THD

The T30 ON and OFF axis.  The Cobreflex horn has quite a wide horizontal dispersion




1w1m on axis RED=T30 ORN=THD; BLK=1824M GREY=THD

Comparison of the T30 and the 1824M, on axis.  Both perform well in their design bandwidth of 250hz to 5khz.  This is due mostly to the Cobreflex horn.  Careful inspection of these curves is similar to those of figs 1, 7 and 8, especially if one considers that these are not gated.

Due to this similarity, it is highly suspected that CLIO is somehow looking at only the frequency being generated and filtering all others.  At this sweep speed, the generator will be at a lower frequency by the time the first reflection of the earlier generated frequency arrives at the microphone.  The time from the speaker to the mic is 2.9mS. Given the incredible clock speeds of microprocessors, this is easy, despite its being mind boggling.  This is all conjecture on my part as CLIO makes no mention of this; it just states that the sweeps are not gated. 

They are incredibly close to gated sweeps on LMS see fig 1




1w1m 350hz mouth aligned pots full CW RED=T30; BLK=1824M

Comparison of the 1824M(BLK) and the T30(RED) on horn. The mid section is first order high pass filtered at 350hz and  first order low pass filtered at 5khz and the tweeter is first order high pass filtered at 5khz.

The 1824M does a better job from 350hz to 1.2khz and is less bright between 1.2khz and 3khz.

The left vertical scale is 3dB per sub division




NOTCH for T30  This was a last minute addendum


The wide band notch (band reject) filter  The design required a 16.8uf capacitor, 0.42mh inductor and a resistor of 4.33W

The capacitor used here is 16.3uf (16) and two inductors of 0.22mh each were used in series.  The 50W potentiometer replaces the fixed resistor to allow the depth of the notch to be adjusted.




The responses of the system.  The filter is wired in series with the T30 as shown in photo 8 below.

The GREY is with the filter bypassed; the pot is full CCW (counterclockwise)

The GREEN is with the pot turned to about 7 o'clock where its resistance is around 5W.

The RED is with the pot at about 8:30 o'clock where its resistance is around 14W

All are measured at 1w1m, ungated

The difference among the 3 curves is about 3dB for each.  The red curve is about 6dB attenuated.  This will be quite noticeable considering the bandwidth from about 800hz to 2.8khz



The schematic with 2 options.  This shows only the mid section of the crossover.  The left optional notch uses a 50W pot and has an optional switch to bypass the notch.  This simplifies bypassing the notch without having to upset the position of the pot.

The right notch option uses a fixed resistor of either 5W or 14W with a bypass switch.

Two resistors could be used in either option if the pot isn't available.  One would be 5W and the other 14W.  A SPST switch would be used to select the desired attenuation. 

The purpose of the pot or two resistors is due to user's preference depending on the music.

NOTE: The right option 2 incorrectly shows the resistor as 5W or 16W; it should be 5W or 14W although it really doesn't matter.; a 16W can be used with no noticeable difference.  During testing, it was noticed that further attenuation of the notch wasn't noticed beyond 14W



Back to the loudspeaker main page

Table of Contents